For me it is a sign of competent design if it does not necessarily require 1% resistors. Even for feedback, what's the deviation from nominal gain with an R 2% off?
Jan
Ticknpop, you are on track about small 1% resistors. They CAN change value dynamically due to instantaneous heating from an audio source (not a sine wave or two). This is something we have known about for about 30 years or more. It is one of the 'secrets' to successful power amp design, to use a larger value low tempco resistor in the feedback network.
It is interesting that with 1% resistors so popular and relatively cheap, that someone would want to use 2% resistors, when that COULD change the gain and throw off any comparison tests that SY might approve of.
It is interesting that with 1% resistors so popular and relatively cheap, that someone would want to use 2% resistors, when that COULD change the gain and throw off any comparison tests that SY might approve of.
Not really a secret. Been part of precision analog teaching since the 80s, in app notes from the likes of AD and natsemi and now championed by Bruce Hofer in his talks.
Forgotten about by audio designers who chase the fad? Possibly in a lot of cases. That or they don't think it matters.
Try getting good CMRR on an input stage with 1% resistors! You have a choice of a transformer, or a whitlock boostrap (or just buy his chip). Not obvious what the competent solution is there...
(note claiming CMRR doesn't matter in domestic audio is not a solution)
Not really, unless the goal is an excuse for avoiding using your ears only.
If one resistor is 2% high and the other is 2% low, then 0.34dB. That's audible, but is worst-case and if the distribution is normal, will only represent a small fraction of the amps. Of course, there's lots of ways around that... assuming engineering. Since 1% are so cheap, that's what I use anyway, and for CMR, I select resistors. For a production amp, there's easier ways if you want to keep CMR or gain tightly controlled.
Let me jus add a comment here. I also have a Denon 3910 universal player I bought in 2002 (I think). At the time, it was much beloved by magazines as an unusually successful universal player. It has 7 DACs inside, from BB, with two in dual differential mode for the L and R channels, the remaininh three catering for SACD/DVD effect channels. It was also one of the first units to feature the Faroudja chipset, so all adjustments of the picture were performed in digital domain. The price was so-so, if memory serves it cost me €1,100, which at the time was something like almost $2,000. A serious machine.
I kept it for its one winning trait - it is unusually linear in any mode I choose.. No uncertainties, no objectionable effects anywhere, it tends to disappear on its own, which I find to be rare indeed in audio. At the time, Marantz had nothing like it, their take on the subject came later on and didn't really excite anyone.
My point is, Denon was a force to be reckoned with even in those days of lingering Marantz fame. And having recently audiotioned theur new players, I was rather underwhelmed by Marantz' offerings, which is why I went for the NAD. It replaced my old Marantz 64 CD player from 1993.
Very nice for the feedback resistor in a big amp - especially CFA where you need low(er) values that result in higher Pd (can be 2-3 watts in a big CFA if you want to preserve SR and bandwidth)
However, I guess your comment was more tongue in cheek
Certainly for small signal duty thin film 1206 SMD devices are an excellent choice.
Naah - seems in volumes of 100 and up those prices are here to stay. They've probably got the high volume metal deposition processes sorted and can churn 'em out for rock bottom prices. Might be that for very high volumes (10's of k and up) thick film is cheaper, but for the stuff I do, these devices are perfectly priced.
I've noticed 0.5W MF seem to be getting more expensive, and suppliers are migrating to the smaller 5-6mm long packages, rather than the 10mm standard. I had to buy some for a project a few weeks ago - they are as much as 10c each up to 25 pcs.
I agree - 1 or 2% should be adequate and definitely so for power amplifiers.
Having said that though, in RIAA EQ's, 1% resistors can give up to a 1.5dB deviation - that's why the trick is to parallel them to improve conformance, rather than spec'ing 0.1% devices.
And for my stuff that's what I do, but if you were designing something to be made in volume a $3 chip would give the accountants a heart attack!
Of course I don't 'need' 80dB CMRR, but it's nice to know it's there and gives all the benefits that come with going balanced (can't use fool cable, XLRs look manly etc).
My understanding is that this has little to do with resistor precision, but rather with source impedance mismatching that occurs mainly at higher frequencies because of inductance and what have you. This is what the Whitlock invention is about. After all, there are more line receivers with laser trimmed resistors on the market, so this is not the distinguishing feature.
Whitlock came up earlier, but now I see it again, I thought it might be good to make that distinction.
Try getting good CMRR on an input stage with 1% resistors! You have a choice of a transformer, or a whitlock boostrap (or just buy his chip). Not obvious what the competent solution is there...
(note claiming CMRR doesn't matter in domestic audio is not a solution)
Yes in specific circuitry like that, better than 1% is probably a good thing. But not as a general rule as is often presented in these pages.
BTW The Hofer trick for the feedback R is to use same resistors for both legs.
For example, if you have a 20k and 1k feedback network, use 20 * 1 k in the series leg. Use same type.
If you do the math you will find that this, to the extend that the resistors are identical, cancels P and V-coefficient modulation.
Jan
- Status
- Not open for further replies.